1232933 狄、發明說明： 【發明所屬之技術領域】 特別是有關於一 本發明係有闕於一種即時診斷專家系統， 種即％式表皮細胞螢光光譜臨床診斷專家系統 【先前技術】 變化測必需經由活體組織切片檢查來偵測細胞的 ;:!、?：片檢查往往需要-至二週的時間，對於患者來 這奴日守間是個難捱的等待。在夂 、 =…易、也最可能及早發現、及早治療，而獲獅。 癌治療率相當高，五年存活率” 7。〜8。%以二 ==貞彳在跳以下，若已㈣端轉移，龍為纖之間。 π 皮膚㉟’若早期發現’都能以簡單的手術或放射療法 :二3皮膚癌經常是三種幾近於良性的基底細胞癌、棘細 二：古黑色素瘤中基底細胞癌很少會發生轉移，棘細 胞在則約有2%是在確診時即已發生轉移，尤其以發生於耳朵、 =恭顳部及枯膜部位者較早發生轉移，而惡性黑色素瘤則很 41 ΐ轉移。皮膚癌的死亡率依治療之早晚，治療時是否有 轉私而疋。基底細胞癌治療後，約只有2%會復發；棘細胞癌， 年生存率約有92°/(> ;而惡性黑色素瘤，依其發現早晚死亡 =相差很大。根據行政院衛生署的統計，在台灣地區口腔癌多 毛生於男性，而皮膚癌則多發生於女性，且口腔癌的死亡率有1232933 D. Description of the invention: [Technical field to which the invention belongs] In particular, the present invention relates to an instant diagnosis expert system, which is a type of epidermal fluorescence spectrum clinical diagnosis expert system [prior art]. Detection of cells by biopsy;:!,?: Examinations often take-up to two weeks, and it is a difficult wait for patients to come to this place. In 夂, = ... Yi, it is most likely to be found early and treated early, and won a lion. The cancer treatment rate is quite high, and the five-year survival rate is "7. ~ 8.%. Two == chastity is below the jump. If the metastasis has metastasized, the dragon is between the fibers. Π Skin can be used if it is found early. Simple surgery or radiation therapy: Skin cancer is often three kinds of almost benign basal cell carcinomas and spines. Second: Basal cell carcinomas in paleomelanoma rarely metastasize, and about 2% of spinal cells are in Metastases have occurred at the time of diagnosis, especially those that occurred earlier in the ears, pheasants, and dysplastic membranes, while malignant melanoma metastases are 41%. The death rate of skin cancer depends on whether the treatment is early or late, whether or not it is treated. There is diversion and relapse. After treatment of basal cell carcinoma, only about 2% will recur; spinal cell carcinoma has an annual survival rate of about 92 ° / (>; and malignant melanoma, depending on its discovery, will sooner or later die = very different. According to the statistics of the Department of Health of the Executive Yuan, oral cancer in Taiwan is mostly hairy in men, while skin cancer is more common in women, and the mortality rate of oral cancer is
越來越南之趨勢Q 另外在美國，根據美國癌症學會2 0 01年所作的報告中指 出’在2001年口腔癌的新增案例約為三萬人，死亡人數約為七 千八百人。皮膚癌的新增案例約為五萬六千人，死亡人數近一 萬人而其中危及生命安全的皮膚癌一黑色素腫瘤在美國的病 1232933 例中正以高比例快速激增（2 〇年内倍數成長）。 由於不管是台灣地 命健康的威脅越來越大於口腔癌與皮膚癌對人民生 ^ ^ ^ ^ ^ 口此對於研發即時、非侵入式臨床 表皮、、，、、哉、、、田I核測系統有迫切的需求。 „至今對於表皮組織的自體勞光量測已有不少的研究鱼專 利，不過習知技術主| bm Ρ 7 J W九”寻 R ·,,. 疋知用早一特性來作辨識，例如Bhaskar 声卜t I 4 #ϋ ’ W及第G1/69199號揭示以特定波長的勞光 =比^«細胞與正常細胞。另外，㈣τ· MeMah〇n等人Increasingly southward trend Q In addition, in the United States, according to a 2001 report of the American Cancer Society, 'the number of new cases of oral cancer in 2001 was about 30,000, and the number of deaths was about 7,800. The number of new cases of skin cancer is about 56,000, and the death toll is nearly 10,000. Among them, life-threatening skin cancer-melanoma is rapidly increasing in a high proportion of 1,332,933 cases in the United States (multiple growth within 20 years). . Because the threat to Taiwan ’s life is more and more greater than the oral and skin cancer threats to people ’s health ^ ^ ^ ^ ^ This is for the development of instant, non-invasive clinical epidermis ,,,,,,,, and There is an urgent need for the system. "So far, there have been many research patents on the measurement of autologous light in epidermal tissues, but the master of the known technology | bm Ρ 7 JW 九" find R · ,,. I know that the earlier characteristic is used for identification, such as Bhaskar sound I 4 # ϋ 'W and No. G1 / 69199 reveal that the labor at a specific wavelength = ratio ^ «cells and normal cells. In addition, ㈣τ · Mahhon et al.
65l74?291 99/45838 f#^^T 程序計算一光譜中數個特定波長的特徵值以推 吊、、且V、增生組織（hyPerPlastic)、腺瘤（adenomatous)或腺癌 (adenocarcmomas) ’如果依照專家系、统的分類來看，此專利所使 用的是假設式（IF · · · THEN)邏輯知識表示法與前向（f㈣㈣)推論 ^、。此外，FrankK〇eni§的美國專利第6,289,236號亦是使用特 ，波長的备光強度比率來推斷發炎組織或癌組織。上述之數種 系統仍有井多難解的問題，因此亟需研發一種即時非侵入式臨 床表皮組織診斷檢測系統。 ％ 【發明内容】 本發明之目的係提供一種即時、非侵入式臨床表皮組織細 胞檢測系統及其方法，以表皮組織細胞之螢光光譜分析來檢測 細胞之變化，使用螢光光譜分析可以檢測細胞是否發生病變、 疋否，、疋細菌感染、是否只是表皮增生、是否癌細胞已經成形、 或癌細胞已經生長到何種情況等資訊。 根據上述目的，本發明之一型態係提供一種即時、非侵入 式g品床表皮組織細胞檢測系統，其包括：一組光纖，其包括第 1232933 :光纖用以弓I導一激發光照射一個體之特定表皮組織，以及第 了光纖用以接收該表皮組織所產生之自體螢光訊號；—、组分光 儀/、匕括第一分光儀用以產生上述激發光，以及第二分光儀 用以純由第二光纖所接收之自體螢光訊號；—光檢測器用以 偵測第二分光儀所接收到的自體螢光訊號；一訊號處理單元， 用以根據該紐測_侧狀自體螢総號計算出該自體螢 光訊=之光譜特性；以及—光错分析單元，其包含—資料庫， 、Λ資料庫與肩自體螢光訊號之光譜特性進行分析以獲得該個 體表皮組織之病變機率或病變種類。 本毛月之另一型態係提供一種即時非侵入式臨床表皮組織 ㈣㈣方法’其包括：將第—分光儀產生之激發光經由第一 光纖引㈣射—個體之特定表皮組織；以第二光、_收該表皮 、戢斤產生之自體*光汛唬至第二分光儀·，以一光檢測器偵測 該自體螢光訊號；根據職檢測輯_到之自體螢光訊號以 一訊號處科元計算出該自«光訊號之光譜特性；以及以一 包含-資料庫的光譜分析單元分析比較該資料庫與該自體勞光 訊號之光譜特性以獲得該個體表皮組織之病變機率或病變種 類。 本發明與習知技術的最大差異點是綜合比較光譜内多個特 性’也就是依照各種組織病變之光譜内的—些特性，如某些特 疋波長的茧光強度比值、特定波長區段的光譜面積值、特定波 峰的上升斜率值等特徵建立一權數表，此權數是依所收隼到的 病變組織㈣料所㈣。本發明採㈣數總合推論法y因此 對於某些特性類似的病變亦可區別，如果依照專家系統」rt System)的分類來看’本發明㈣算法類似㈣式知識表示法加 上以機率為基礎的推論法，因此在學理上與先前技術的方法是 1232933 【實施方式】 本發明係先建立一套表皮細胞螢光光譜資料庫，藉此建立 一即時性、非侵入式之表皮組織細胞檢測系統及方法。臨床上 使用時，醫師馬上可以得知組織細胞之情況，若系統判斷為癌 細胞’則再進行切片檢查作二次確認，如此不僅可以大符降低 醫療資源、節省醫師與病患等待檢驗結果之時間，更可以在最 ^夬日守間i現病源加以治療。本發明之檢測糸統係以口腔癌與皮 膚癌為主要應用，因為此兩種癌症之表皮組織相當容易取得螢 光光譜，而且也已有臨床研究，可幫助本發明之檢測系統快速 建立赏光光瑨資料庫。本發明之具體實施方法將詳述如下。 1 螢光量測設備建置 本兔明之檢測系統裝置如第1A圖之示意圖與第1B圖之照 所=，包含一個光源1，用來產生激發光線；一組分光儀，其 St光光儀2(或稱第-分光儀)，用來產生特定波長之 ^定^ ]為接收光之分光儀4(或稱第二分域），用來接 所示之檢測様。s / 處如可放置第⑺圖 光昭…:：或一組光纖，如第5圖所示，用來引導激 先…、射表皮組織與接收表皮組織產 文 5,用來接收自體瑩光訊號;以 -，測 電腦7,利用所I於μ Α 4工制用电細6與一數據分: 所接收到的自體螢光光譜，#過皆曾、 “的特性辨識出組織病變。 '工仏法運异，可彳 2· f家系統建構演算法設計 考慮-般V輯了:二，合型’因此在演算法的設計上不们 Ϊ醫學專家系、=:=法與推論方法，本發明人等嶋 鼻法之設計，並且 \ =機率的模式為基礎來進行沒 &月之濟异法攻好能 月匕同日卞根據多項光譜# 1232933 特徵來進行設計。 / 第2圖係演算法之示意圖，在圖中，ρι、p2 ··. p7指的是 _Perty，就是光譜的各項性質，此處的1、2、···、7只是一個 代表編號，例如P1為面積比（area ratio)、P2為螢光強度比 ratl〇)、P3 為斜率（sl〇pe)、…·Ρ7 為面積比 2(area ratio ^ ^上本發明之演算法具有擴充性質數量的彈性，意即熟 而n勢人士可以根據實際需求自行定義出光譜的各項特性。 而 Dl、D2、m τλ 管 …X扣的是病變（Crease)或症狀種類，此渖 …二不：定二辨識的病變數與種類，可以依實際需求而定義： 譜-定：有之：异f上的核心觀念為：不同病變(dis叫組織的光 會有重 時，只比較-種光譜=不、=:_:’如正_ 專利所著重的地方，但是者二 &也疋許多論文與 如果還是使用—種光=田有二種以上的病變組織要比較時， 會變的較複雜，如果選不同點作比較，此工作就 複雜性又提高不少，而本 ;特性一起作判斷， 本空間的建立，不同光丄二 =率為基礎，藉由樣 演算法的核心資料庫，本―艾之間相對應的機率構成此 病變的機率，用直觀的二可光譜特性對應於 可咸性越大，由於機率是统 向、示成為此病變的 是符合臨床上的判斷模式、，、果/、要樣本空間足夠大， 性，都可以由本演算法推論出大對^於某種组織的光譜特 可以表示出該組織為某病變的可能機率。丙夂的機率值，此機率 如$述，本發明之演算法 ；' 相對應的機率，因此可^^為不同光譜特性與病變之間 (W軌table)」，權數表的=機率:的集合建構成-「權數表 建構示意圖如第3圖所示。第3圖 10 1232933 貝料庫（database)中存放的是各種病變組織的特性表，經過 =十《後可以得到本發明之演算法的核心權數表。以下以數 +式5兒明本發明之演算法。 久本發明之演算法的資料庫將包含數個表格，這些表格 不同的定義。首先是每一種病變是一個大類別，有苴獨 =’此處定義為D(取Disease之意)，因此病變種類資料 犀的樣本空間為： 乃-，其中 yte# .....(1) ::A，…名表示各自獨立的表格’其中7刀，…,々表示病變的 ^ > ^ ^'J Disease 1 . Disease 2 > Disease 3 > Disease k # ^ 足疋由使用者所定義的。 在母個病變表中，是許多組光譜特性的集合，此處定義為^， 其中樣本空間的表示法如下： ，其中 (2) = f不布林(BO01，)值，表示編號仏病變中的第/個樣本 ’心；弟Z項光譜特性的布林值。 羞在—開始建構資料庫時，病變或症狀的種類與名稱已事先定 以將2所使用的樣本也是必需已知是屬於那種病變組織才可 爾到資料庫中。舉例來說，欲辨識皮膚組織中的基底 、、、田月已癌、棘細胞癌、亞降 a., 織的光ml 黑色素瘤、牛皮癖、疲、等五種不同組 6 =其各自的資料庫編號就依序為A，A，D3，Z)4，Z)5。在定義 ί種Γί或症狀種類與名稱之後便可將樣本依不同的病變或症 狀種類建立各自的資料庫Α。 1上每個資料庫Μ，利用生物統計的數學方法可以得到各光 可由以下的數學式表示： 1232933 心=則A—)，其中《心ΛΑ(3) 其中心表示病變々的笙 ^ ^ , 乐”種光譜特性的統計機率值"丨 建立各自病變組織的 手值户（I )。當 4 #叮 ' 5先5¥4寸性資料庫之各特性的 後，便可以建構本演首 t丨扪..死冲機率值之 t法的核心權數表，以p (wel2hn# - ^ω，其中/aeyV…⑷ 表不。 式⑷:P權數表鳴本空間表示式，係由以 。 ^ ^是根據已有的樣本數所建立的資料庫與權數表， 此凟异法根據知識庫的 I忉以卜成明 法。 内奋推确出未知病變組織所屬病變的方 批座虽一個新的樣品組織的光譜特性產生時，首先合產生-個 對應到各光言善特性布林值陣列，以表示，其陣列如曰下·· 久=fc} ’ 其中 ......(5) 第·個弁\兑 病艾名％’ 6/表不此未知病變組織的光譜特性中 弟i们光ΰ晋特性的布林值。 數^ t生之後’便可以根據權數表進行推論，根據a與權 數表#，我們可以得出推論式如下： ηThe 65l74? 291 99/45838 f # ^^ T program calculates the characteristic values of several specific wavelengths in a spectrum to suspend, and V, hyperperplastic, adenoma (adenomatous), or adenocarcmomas. Looking at the classification of expert systems and systems, this patent uses a hypothetical (IF · · · THEN) logical knowledge representation and forward (f㈣㈣) inference ^ ,. In addition, US Patent No. 6,289,236 by Frank Koeni also uses a special wavelength ratio of prepared light intensity to infer inflammatory or cancerous tissue. The above-mentioned several systems still have many problems, so it is urgent to develop an instant non-invasive clinical epidermal tissue diagnosis and detection system. [Summary of the invention] The object of the present invention is to provide an instant, non-invasive clinical epidermal tissue cell detection system and method, which uses fluorescence spectrum analysis of epidermal tissue cells to detect changes in cells, and uses fluorescence spectrum analysis to detect cells. Information such as whether a disease has occurred, whether it is, bacterial infection, whether it is only epidermal hyperplasia, whether cancer cells have formed, or how cancer cells have grown. According to the above purpose, one aspect of the present invention is to provide an instant, non-invasive epidermal tissue cell detection system for g product bed, comprising: a set of optical fibers, including the 1232933: the optical fiber is used to guide an excitation light to The specific epidermal tissue of the individual, and the first optical fiber to receive the autofluorescence signal generated by the epidermal tissue;-, a component spectrometer /, a first spectrometer to generate the above-mentioned excitation light, and a second spectrometer For purely self-fluorescent signals received by the second optical fiber;-photodetector for detecting self-fluorescent signals received by the second spectrometer; a signal processing unit for measuring the side according to the button The autofluorescence signal calculates the spectral characteristics of the autofluorescence signal; and-the optical error analysis unit, which includes-a database, a Λ database, and the spectral characteristics of the autofluorescence signal of the shoulder to analyze to obtain the Probability or type of lesion in individual epidermal tissue. Another aspect of the present month is to provide an instant non-invasive clinical epidermal tissue method, which includes: irradiating the excitation light generated by the first spectrometer through a first optical fiber to a specific epidermal tissue of the individual; Light, _ collected the epidermis, self-generated * light to the second spectrometer, and detected the auto-fluorescence signal with a light detector; according to the job detection series _ to the auto-fluorescence signal Calculate the spectral characteristics of the «optical signal with a signal element; and analyze and compare the spectral characteristics of the database and the autologous optical signal with an include-database spectral analysis unit to obtain the individual epidermal tissue. Probability or type of lesion. The biggest difference between the present invention and the conventional technology is the comprehensive comparison of multiple characteristics in the spectrum, that is, according to the characteristics of various tissue lesions in the spectrum, such as the cocoon light intensity ratio of certain special wavelengths, and the specific wavelength range. Spectral area values, rising slope values of specific peaks, and other characteristics establish a weight table, and this weight is based on the collected lesion tissue data. The inference method y of the present invention can therefore be distinguished for certain lesions with similar characteristics. If you look at the classification according to the "rt system" of the expert system, the "invention" algorithm is similar to the knowledge representation method plus the probability Basic inference method, so the method of science and the prior art is 1232933 [Embodiment] The present invention first establishes a set of epidermal cell fluorescence spectrum database, thereby establishing an instant, non-invasive detection of epidermal tissue cells System and method. In clinical use, the physician can immediately know the status of the tissue cells. If the system determines that the cells are cancerous, then the biopsy will be performed for secondary confirmation. This will not only greatly reduce medical resources, save doctors and patients waiting for test results. Time, can be treated at the source of the disease at the last day. The detection system of the present invention uses oral cancer and skin cancer as the main applications, because the epidermal tissues of these two cancers are relatively easy to obtain fluorescence spectra, and clinical studies have also been conducted to help the detection system of the present invention to quickly establish the light瑨 Database. The specific implementation method of the present invention will be described in detail as follows. 1 Fluorescence measurement equipment The detection system of this rabbit is built as shown in the schematic diagram in Figure 1A and the photo in Figure 1B. It contains a light source 1 for generating excitation light; a component photometer, its St photometer 2 (or the first-spectrum spectrometer), which is used to generate a specific wavelength of the spectrometer 4] is the spectrometer 4 (or the second spectroscopic field) that receives light, and is used to connect to the detection detector shown. The s / place can be placed in the first picture Guangzhao ... :: or a group of optical fibers, as shown in Figure 5, used to guide the radical ..., shoot epidermal tissue and receive epidermal tissue Article 5, used to receive autofluorescence The signal is:-, test computer 7, using the 6 and 1 data points of the μA4 working system: the received autofluorescence spectrum, #pass all, "The characteristics identify tissue lesions. 'The method of operation is different, but the design considerations of the system construction algorithm of 2f family-general V series: two, fit type' Therefore, the design of the algorithm is not based on the medical expert department, =: = method and inference method , The inventors and others design the nose-brush method, and the \ = probability model is used as the basis to carry out the & moon's magic method to attack the good moon dagger and the sundial according to the characteristics of multiple spectrum # 1232933 to design. / Figure 2 A schematic diagram of the algorithm. In the figure, ρ, p2 ··· p7 refers to _Perty, which is the properties of the spectrum. Here, 1, 2, ..., 7 are just a representative number. For example, P1 is Area ratio, P2 is the fluorescence intensity ratio ratl0), P3 is the slope (slOpe), ... · P7 is the area ratio 2 (area ratio ^ ^ The algorithm of the present invention has the flexibility of expanding the number of properties, which means that people of n potential can define the characteristics of the spectrum according to actual needs. The Dl, D2, m τλ tubes ... ) Or the type of symptoms, this ... No: The number and type of lesions identified by Ding 2 can be defined according to actual needs: Spectrum-Ding: There is: The core concept on different f is: different lesions (dis called tissue light When there will be heavy, only compare-Spectral = No, =: _: 'Ru Zheng _ The place where the patent is focused, but the second & also many papers and if still used-Kind of light = Tian has more than two kinds of Lesion tissues will become more complicated when compared. If different points are selected for comparison, the complexity of this work will increase a lot, and the characteristics will be judged together. The establishment of this space will be based on different light sources. With the core database of the sample algorithm, the corresponding probability between Ben and Ai constitutes the probability of this lesion, and the intuitive binary spectrum characteristic corresponds to the greater the saltiness. Since the probability is uniform, it becomes this The disease is in line with clinical judgment The model, ,, fruit /, and the sample space are large enough, and the sex can be deduced by this algorithm. The spectrum of a certain tissue can be used to indicate the probability of the tissue as a certain disease. This probability is as described in the algorithm of the present invention; the corresponding probability, so it can be ^^ between different spectral characteristics and lesions (W-track table) ", the weight table's = probability: a set of constructs-" The weight table construction diagram is shown in Fig. 3. Fig. 10 1232933 The database stores the characteristic tables of various diseased tissues, and the core weight table of the algorithm of the present invention can be obtained after ten seconds. The algorithm of the present invention will be described by the following formula +5. The database of algorithms of the present invention will contain several tables, these tables having different definitions. The first is that each type of lesion is a large category, and there is uniqueness = 'Here it is defined as D (takes the meaning of Disease), so the sample space of the type of lesion data is: Nai, where yte # ..... (1 ) :: A, ... names represent separate tables 'of which 7 knives, ..., 々 represent lesions ^ > ^ ^' J Disease 1. Disease 2 > Disease 3 > Disease k # ^ is enough by the user As defined. In the mother lesion table, it is a collection of many sets of spectral characteristics, which is defined here as ^, where the representation of the sample space is as follows:, where (2) = f 不 布林 (BO01,) value, which represents the number 仏 in the lesion The / sample of the 'heart; the Bollinger value of the spectral characteristics of the Z term. Ashamed—At the beginning of the construction of the database, the type and name of the disease or symptom have been determined in advance. 2 The sample used must also be known to belong to that type of disease tissue before it can be included in the database. For example, to identify five different groups of basal, skin, cancer, spinal cell carcinoma, sub-a.a., Melanoma, psoriasis, fatigue, etc. in the skin tissue 6 = their respective The database numbers are sequentially A, A, D3, Z) 4, Z) 5. After defining Γ, or the type and name of the symptom, the sample can be set up as a separate database A according to different types of disease or symptom. Each database M on 1 can be obtained using the mathematical method of biostatistics. Each light can be expressed by the following mathematical formula: 1232933 Heart = 则 A —), where "心 ΛΑ (3) whose center represents the pathological change ^ ^, Statistical Probability Values of Spectral Characteristics of Music "" 丨 Establishing Hand-Valued Accounts (I) of the respective diseased tissues. After 4 #ding '5 first 5 ¥ 4 inch sex database, you can construct the first performance t 丨 扪 .. The core weight table of the t-method for the value of the probability of the dead punch, expressed by p (wel2hn #-^ ω, where /aeyV...⑷ is not expressed. Equation ⑷: P weight table Naomi space expression. ^ ^ Is a database and weight table created based on the number of existing samples. This method is based on the I method of the knowledge base. Inner endeavors have determined that the square of the unknown lesion tissue belongs to a new one. When the spectral characteristics of the sample tissue are generated, an array of Bollinger values corresponding to each optical-speech characteristic is first generated to indicate that the array is as follows: · Jiu = fc} 'where ... (5) The first one \ 's disease name%' 6 / expresses the characteristics of the unknown characteristics of this unknown lesion tissue . ^ After the number of students t 'can be deduced according to the weights table, according to the number of weights and a table #, we can draw the following inferences formula: η
Tk =Σ^α|έ/ =true) -----(6) 率她果係由'決^，7^示的是&對應於第〜病變中的機 :左’ §對應於某種病變之機率總值越高表示屬於該種病變 之機率越大，因此可以推論出'屬於那—種病變的可能性最大。 另外本演异法還包括-自動權數修正演算法，所謂的自動 榷數修正示意圖如第4圖所示。在圖中，當推論結以已知之 後，便心新增到第W固資料庫中，因此權數表"的辑 自動被修正了。 雖然本發明之檢測系統係針對量測細胞組織的自體榮光光 12 1232933 譜來辨識病變種類，事實上本發明檢測 識自體螢丼之伞咬4±以…、”之’/、异法除了可辨 r a先糾性外，對於特性種類並不限| 辨t,J: 行定義出光譜的各項特性種類。而且t B '沩芰或症狀數目與種類並不限自 ° 與各種光譜特性生地自行定義並 為基礎，藉發明檢測系統之演算法以機率 應的機率構成此、、宫曾本M k °曰特性/、病受之間相對 光譜特性2 : 貧料庫’本演算法可顯示出各種 _ 子應於病蜒的機率，用直觀的觀點來看，機率艘古矣 不成為此病變的可能性越大，由於機率是統計的 /、 本空間足夠大，Η锌人浐广 、’、口果’只要樣 大疋付5臣品床上的判斷模式，因此對於任付一插 組織的光譜特性，都可以由本演算 甘Η7 一種 率值。“认士一、 不一套推-出對應於某病變的機 ;㉝异法的推論結果是一個機率總合的數值， 此可以看出新樣本對應到各已知病變組織的機率值，機率她人 值越大表示該病變的可能性越大，但是本演算法還提供了可^ 為其他病變的可能性的機率值供作參考，這對於臨床的判斷2 是提供了更多的資訊讓醫護人員參考’與以往類似的光削寺徵 辨識結果非正即負的資訊有相當大的差別。 此外，一般辨識病變組織螢光光譜的方法都是直接以28〇 nm波長的紫外光源照射人體組織以取得螢光光譜，因為從以往 的相關研究指出人體組織内可發出自體螢光的物質，例如，蛋 白質如彈力蛋白（elastm)，胺基酸如色胺酸、酪胺酸或苯丙胺 酸，嘌呤（purmes)如腺嘌呤（adenme)或鳥糞嘌呤（guanme)，嘧 啶，核酸如腺苷、鳥糞嘌呤核苷、DNA或RNA，這些物質吸收 280 nm波長左右的紫外光後，其自體螢光光譜會在34〇nm〜39〇 nm附近產生波峰（peaks)。不過對於細胞内單一物質如不同胺 基酸的自體螢光光譜的特性這類的相關研究卻很少見，然而 13 1232933 種4寸性可能與疾病狀態相關，例如癌症組織在螢光光譜上與正 常組織的螢光光譜之間的不同點可能即是細胞内不同胺基酸之 里所產生’因此本發明著重於此更深入的探討，以不同濃度的 胺基酸來建立細胞内組成成份之光譜資料庫，並且亦驗證光譜 辨識之演算法’因為本演算法的最大特色就是不限定A資料庫 的種類’仏可以是病變名稱，當然也可以是胺基酸的各種不同 濃度或類型。 本發明之實施例主要採用病理細胞株培養的方法來代替活 體病人細胞，因為對組織細胞而言，不論是培養的或是從病人 身上取出的，其細胞組成並沒有差異，因此同樣可以用來取得籲 勞光量測光譜，另一方面為驗證光譜辨識之演算法，以不同濃 度的胺基酸來建立細胞内組成成份之光譜資料庫。 本發明之實施亦考量到臨床使用的安全性，由於目前以組 、、哉的自體螢光來辨識癌細胞的方法都是以短波長的紫外光來照 $而糸外光會破壞細胞物質，甚至有致癌的危險性，並不適 床上使用，因此本發明之另一目的是尋找更適合的非破壞 =放發光的自體螢光光譜特性來區別病變組織與正常組織。本 發明採用的螢光光譜量測設備可自行調整入射光的波長，從《工· 外光到可見光到紫外光等全波域，並且發現採用綠光就可以 到所需的螢光光譜。 —以下以具體貫施例詳細說明本發明，然而本發明並不限於 實施例所揭示之範疇。 、 實施例1實際量測人體表皮組織之螢光光譜 以第1B圖所示之裝置， 5圖所示之一組光纖，以如第 表皮組織之自體螢光光譜，上 ，並將檢測樣品放置處3改使用如第 ;6圖所示方法直接量測正常志願者 以波長5〇〇nm的綠光為光源，掃插 14 1232933 光言番從5 1 Onm至60Onm，測試3位正常志願者之結果如第7A〜7C 圖所示，結果發現在波長544.6nm的位置有一波峰，並證明皮 膚組織以綠光作入射光也可以產生自體螢光光譜，這是一個以 往文獻中沒有記載的資訊。 實施例2量測不同胺基酸濃度之螢光光譜 以第1B圖所示之裝置，測試不同胺基酸濃度之光譜，結果 如第 8A〜8D 圖所示。第 8A圖是酪胺酸（tyrosine)濃度 0.05mg/ml，以300 nm波長光源照射所得到的螢光光譜；第 8B〜8D圖是苯丙胺酸（phenyl alanine)濃度0.005mg/ml以及酷胺 酸（tyrosine)濃度0.05mg/ml之結果，其中第8A圖之照射光源波 長為300nm，掃描範圍為310nm〜580nm，第8C圖之照射光源 波長為320nm，掃描範圍為330〜620nm，第8D圖之照射光源 波長為320nm，掃描範圍是325〜620nm。 由以上結果顯示不同濃度的混合溶液在螢光光譜上會顯示 出不同的波峰，以此為基礎，可建立含有不同濃度胺基酸之細 胞内組成成份的光譜資料庫。 實施例3量測不同培養細胞株之螢光光譜 同樣以第1B圖所示之裝置，測試不同培養細胞胞株之光 譜。以肝癌細胞與黑色素細胞為例，採用波長280 nm之紫外光 照射，所得到的光譜圖如第9A〜9B圖。 在第9A圖中，曲線由上而下分別為PBS，PBS+肝癌細胞， PBS +黑色素瘤細胞，入射光源為波長280nm的紫外光，掃描 範圍為290nm至540nm。在第9B圖中，曲線由上而下分別為 PBS，PBS +黑色素瘤細胞，PBS +肝癌細胞，入射光源為波長 420nm的紫光，掃描範圍為440nm至820nm。其中PBS指的是 培養JHI溶液。結果顯示，可以不同掃描範圍分別腫瘤細胞。 15 1232933 目如越來越多的 癌病變，所庳用沾西子研九夤試以光學量測方法來檢測組織 發鸯光^、原理有光的散射、雷射反應、波長改變、自 特性计 寺寺。由本發明實驗發現，不僅是自鲈狄氺 ”生，其他的光學特性 疋自體以 應用於本發明之演算法中。用於_癌細胞組織，而且可以 雖然本發明已以較佳實施例 本發明，任何熟習此技藝者，在不士 ’、，、，、亚非用以限定 An" 在不脫離本發明之精神釦r同肉， 當可作各種之更動與潤飾，因此 ^申和乾圍内 申請專利範圍所界定者為準。 X之保σ又靶圍當視後附之Tk = Σ ^ α | έ / = true) ----- (6) The rate of her fruit is determined by 'decision ^, 7 ^ is & corresponds to the machine in the ~ lesion: left' § corresponds to a certain The higher the total probability of a lesion, the greater the probability of belonging to that lesion, so it can be inferred that 'belonging to that lesion is most likely. In addition, this algorithm also includes an automatic weight correction algorithm. The so-called automatic number correction diagram is shown in Figure 4. In the figure, when the inference is known, it is added to the W-th database, so the compilation of the weight table " is automatically corrected. Although the detection system of the present invention is aimed at measuring the autonomic glory light 12 1232933 spectrum of cell tissues to identify the type of lesion, in fact, the present invention detects and recognizes the umbrella bites of autofluorescent crickets 4 ± by “,” of “/, different methods In addition to distinguishing the correctness of ra, there is no limitation on the type of characteristics | The t, J: line defines each characteristic type of the spectrum. And t B '沩 芰 or the number and type of symptoms are not limited to ° and various spectra The characteristics of habitats are defined by themselves and based on the algorithm that invented the detection system to construct this, with the probability of the probability, Miyazaki M k ° said characteristics /, relative spectral characteristics between illness and disease 2: poor material library 'this algorithm Can show various probabilities that the disease should be sick. From an intuitive point of view, the greater the probability that the protagonist does not become this disease, since the probability is statistical, the space is large enough, As long as Guang, ', and mouth fruit' have the same judgment model on the bed, the spectral characteristics of any intervening tissue can be calculated by this calculation. The value of Gan 7 can be calculated by this calculation. -Explain the machine corresponding to a certain disease; surprise method The inference result is a sum of probability values. It can be seen that the new sample corresponds to the probability value of each known disease tissue. The greater the probability of her, the greater the probability of the disease, but the algorithm also provides ^ It is used as a reference for the probability of the possibility of other lesions. This is for clinical judgment. 2 It provides more information for medical staff to refer to the information that is similar to the previous positive or negative results of the Guangsha Temple sign. Big difference. In addition, the general method for identifying the fluorescence spectrum of diseased tissues is to directly irradiate human tissues with an ultraviolet light source at a wavelength of 28 nm to obtain a fluorescence spectrum, because previous related studies have pointed out that substances that can emit autofluorescence in human tissues, For example, proteins such as elastm, amino acids such as tryptophan, tyrosine, or phenylalanine, purmes such as adenme or guanme, pyrimidines, nucleic acids such as adenosine, Guanosine nucleosides, DNA or RNA, these substances absorb ultraviolet light at a wavelength of about 280 nm, and their autofluorescence spectrum will generate peaks around 34nm ~ 39nm. However, related research on the characteristics of the autofluorescence spectrum of a single substance in the cell, such as different amino acids, is rare. However, 13 1232933 four-inch sex may be related to the disease state, such as the fluorescence spectrum of cancer tissues. The difference from the fluorescence spectrum of normal tissues may be generated by different amino acids in the cell. Therefore, the present invention focuses on this in-depth discussion, and uses different concentrations of amino acids to establish intracellular components. The spectrum database and the algorithm for verifying the spectrum identification 'because the biggest feature of this algorithm is that the type of the A database is not limited'. It can be the name of the lesion, or it can be various concentrations or types of amino acids. The embodiment of the present invention mainly adopts a method of culturing a pathological cell line to replace a living patient cell, because for a tissue cell, whether it is cultured or taken from a patient, there is no difference in cell composition, so it can also be used for Obtain the light measurement spectrum. On the other hand, in order to verify the algorithm of spectral identification, different concentrations of amino acids are used to establish a spectral database of intracellular components. The implementation of the present invention also takes into account the safety of clinical use. Because the current methods of identifying cancer cells with autofluorescence of groups, and plutonium are irradiated with short-wavelength ultraviolet light, external light will destroy cellular material. There is even a risk of carcinogenesis and it is not suitable for bed use. Therefore, another object of the present invention is to find a more suitable non-destructive = radioactive autofluorescence spectral characteristic to distinguish diseased tissue from normal tissue. The fluorescence spectrum measuring device used by the present invention can adjust the wavelength of incident light by itself, from the full wave range of "work and external light to visible light to ultraviolet light," and found that the green light can be used to obtain the required fluorescence spectrum. -The present invention will be described in detail through specific examples, but the present invention is not limited to the scope disclosed in the examples. Example 1 The fluorescence spectrum of human epidermal tissue was actually measured with the device shown in Fig. 1B and a set of optical fibers shown in Fig. 5 was taken with the autofluorescence spectrum of epidermal tissue. Place 3 and use the method shown in Figure 6 to directly measure normal volunteers. Use green light with a wavelength of 500 nm as the light source. Sweep 14 1232933 Guangyanfan from 5 1 Onm to 60 Onm. Test 3 normal volunteers. The results are shown in Figures 7A to 7C. The results show that there is a peak at the wavelength of 544.6nm, and it is proved that the skin tissue can generate autofluorescence spectrum with green light as the incident light. This is not recorded in previous literature. Information. Example 2 Measurement of fluorescence spectra of different amino acid concentrations Using the apparatus shown in Figure 1B, the spectra of different amino acid concentrations were tested, and the results are shown in Figures 8A to 8D. Figure 8A is a fluorescence spectrum obtained with a tyrosine concentration of 0.05 mg / ml and irradiated with a light source at a wavelength of 300 nm; and Figures 8B to 8D are a phenyl alanine concentration of 0.005 mg / ml and a phosphoric acid (Tyrosine) Concentration 0.05mg / ml, where the wavelength of the irradiation light source in Figure 8A is 300nm, the scanning range is 310nm ~ 580nm, the wavelength of the irradiation light source in Figure 8C is 320nm, the scanning range is 330 ~ 620nm, and in Figure 8D The wavelength of the irradiation light source is 320nm, and the scanning range is 325 ~ 620nm. The above results show that mixed solutions of different concentrations will show different peaks on the fluorescence spectrum. Based on this, a spectral database of intracellular components containing amino acids of different concentrations can be established. Example 3 Measurement of the fluorescence spectra of different cultured cell lines. Also using the apparatus shown in Figure 1B, the spectra of different cultured cell lines were tested. Taking hepatocellular carcinoma cells and melanocytes as examples, the ultraviolet spectrum was irradiated with a wavelength of 280 nm, and the obtained spectrum diagrams are shown in Figs. 9A-9B. In Fig. 9A, the curves are PBS, PBS + liver cancer cells, PBS + melanoma cells from top to bottom. The incident light source is ultraviolet light with a wavelength of 280 nm, and the scanning range is 290 nm to 540 nm. In Figure 9B, the curves are PBS, PBS + melanoma cells, PBS + liver cancer cells from top to bottom. The incident light source is purple light with a wavelength of 420 nm, and the scanning range is 440 nm to 820 nm. Where PBS refers to the cultured JHI solution. The results show that tumor cells can be scanned separately for different scanning ranges. 15 1232933 As more and more cancerous lesions appear, I used Zanxiziyan Jiu Ji to test the optical emission of tissues with optical measurement method. Principles include light scattering, laser response, wavelength change, and self-characteristics. Temple Temple. It is found from the experiments of the present invention that not only is it from the perch, but other optical characteristics are used in the algorithm of the present invention. It is used for cancer cell tissue, and although the present invention has been described in a preferred embodiment, Invention, anyone who is familiar with this skill is used to define An " in Brazil, ",,,,, and A " without deviating from the spirit of the present invention. It can be used for various modifications and retouching. Therefore, Shen Heqian The definition of the scope of the patent application within the scope shall prevail. The guarantee of X and the target scope shall be attached as follows.
16 1232933 【圖式簡單說明】 二第1A圖顯示本發明之即時非侵入式表皮細胞螢光光譜臨 床移斷檢測糸統；第1B圖顯示上述系統之具體實施裝置的照 第“圖頒不本發明之即時非侵入式表皮細胞螢光光譜臨床 診斷檢測系統所採用的演算法。 第3圖顯不本發明之即時非侵入式表皮細胞螢光光譜臨床 診斷檢測系統中權數表之建構示意圖。16 1232933 [Schematic description] Figure 2A shows the instant non-invasive epidermal cell fluorescence spectrometry detection system of the present invention; Figure 1B shows a photo of the specific implementation device of the system described above The algorithm used in the invention of the instant non-invasive epidermal cell fluorescence spectrum clinical diagnostic detection system. Figure 3 shows a schematic diagram of the construction of the weight table in the instant non-invasive epidermal cell fluorescence spectrum clinical diagnostic detection system of the present invention.
第4圖顯示本發明之即時非侵入式表皮細胞螢光光譜臨床 診斷檢測系統中自動權數修正示意圖。 第5圖頒不本發明之實施例採用光纖的情形。 第6圖顯示本發明實施例1中使用光纖實際量測人類表皮 組織螢光光譜·之情形。 第7A〜7C圖顯示不同正f志願者表皮組織榮光光譜；第7 圖係編號1之志願者之夯古益么士 s @ m〆 只可 < 尤σ曰結果，弟7Β圖係編號2之志願者έ 光譜結果’第7C圖係編號3之志願者的光譜結果。 第8Α,圖顯示不同胺基酸濃度之榮光光譜；第8α^ 絡胺酸之光言晋結果，第8R〜m μ 圖係路胺酸與苯丙胺酸之光言並|FIG. 4 shows a schematic diagram of automatic weight correction in the instant non-invasive epidermal cell fluorescence spectrum clinical diagnosis detection system of the present invention. FIG. 5 illustrates a case where an optical fiber is not used in the embodiment of the present invention. Fig. 6 shows the actual measurement of the fluorescence spectrum of human epidermal tissue using an optical fiber in Example 1 of the present invention. Figures 7A to 7C show the glory spectra of the epidermal tissues of different volunteers; Figure 7 is the volunteer's Gu Gu Mo Shi No. 1 s @ m〆 只 可 < You σ said the result, brother 7B Figure No. 2 Volunteer's spectrum results' Figure 7C is the spectrum results of volunteer number 3. Figure 8A, the photo shows the glory spectrum of different amino acid concentrations; the results of the 8α ^ complex amino acid photo-synthesis, and the 8R ~ m μ figure is the photo-synthesis of glutamic acid and phenylalanine |
果，其中第8Β圖係300nm之入射古、塔曰、 之入射先源，31〇〜580nm之掃描』 圍，第8C圖係320nm之入射光 — . 耵九原330〜620nm之掃描範圍 弟8D圖係320nm之入射光源， 和尤嫁，325〜620nm之掃描範圍。 第9A〜9B圖顯示不同土立羞^ ，The result is that the 8B picture is the 300nm incident ancient, tower, and incident source, and the scan is from 31 to 580nm. The 8C picture is 320nm incident light—. .Kiuhara 330 ~ 620nm scanning range. 8D The picture shows the incident light source at 320nm, and the scanning range of 325 ~ 620nm. Figures 9A ~ 9B show different local shame ^,
个口養細胞株之光譜結果；第9A 280nm之入射光源，290〜54〇nm tθ π ° nm之輙描靶圍，第9β圖係42〇]η 之入射光源’ 440〜820nm之掃描範圍。 【符號說明】 1〜光源， 2〜激發光之分光儀（第1分光儀） 17 1232933 3〜檢測樣品放置處， 4〜接收光之分光儀（第2分光儀）； 5〜光檢測器； 6〜控制用電腦； ’ 7〜數據分析用電腦， 8〜檢測樣品， 9〜光纖； E〜激發光； F〜自體螢光訊號。 •Spectral results of individual oral cell lines; 9A 280nm incident light source, 290 ~ 54nm tθ π ° nm trace target range, and 9β figure is 42 °] incident light source 'scanning range of 440 ~ 820nm. [Symbol description] 1 ~ light source, 2 ~ excitation spectrometer (first spectrometer) 17 1232933 3 ~ place for detecting sample, 4 ~ receiving light spectrometer (second spectrometer); 5 ~ light detector; 6 ~ control computer; '7 ~ data analysis computer, 8 ~ detection sample, 9 ~ optical fiber; E ~ excitation light; F ~ autofluorescence signal. •
1818